We found that 20 (77%) interns had 80 episodes of continuous duty hour reporting greater than 30 hours. The intern factor associated with reported noncompliance was the number of inpatient rotations (a strong inverse association). Noncompliance varied significantly among interns after accounting for the workload factors. The workload factors associated with reported noncompliance were the number of Non-hospitalist patients and increasing total patients (modest positive associations). The number of admissions on-call, number of admissions after midnight, total patients post-call and number of discharges post-call were not found to be significant risk factors for noncompliance.
We designed this study to inform the modification of intern workflow to ensure compliance with the 2003 ACGME duty hour requirements in the U.S. Based on the preliminary analysis of this data, the pediatric residency program decreased the cap on overnight admission in July 2010 from 8 to 6 patients per intern for the first 2 call nights of the intern’s first ward rotation. The ACGME, however, has eliminated extended duty periods for interns, effective July 1, 2011. The residency program then reduced the maximum number of patients each intern could round on each day during the first week of his/her first ward rotation. The housestaff has not, however, reported this change to be helpful. While residents beyond their intern year are permitted to take overnight call, their workflow can differ from that of the interns. The method and results of this study, nonetheless, have important implications for this revised system and perhaps similar systems in other countries.
Residents will still need to comply with duty hour restrictions and residency programs will need to determine what factors contribute to noncompliance. There is limited literature on this topic. This may be due, in part, to the fact that collecting data on resident workload is difficult, especially if it is dependent on residents maintaining patient logs. Data was available for this study from software designed, in part, to facilitate resident hand-offs, which is integrated into their workflow . Many residency programs, including a recent large-scale collaborative research study [12, 13], are focusing on hand-off communication as a way to decrease medical errors that may result from the increased number of hand-offs. Data systems to facilitate hand-off communication could be developed to capture granular data, which could be used for other purposes such as studying factors associated with noncompliance.
The identification of specific intern skills and attributes associated with compliance is a very important area of future study. Our results demonstrate that experience was associated with a larger reduction in the risk of noncompliance than other system factors. Noncompliance also varied significantly among the individual interns after accounting for the workload factors. These skills and attributes may not, however, be related to cognitive knowledge but skills such as organization and efficiency. There is, unfortunately, limited literature on resident productivity and the available literature is mainly focused on the emergency department . If specific skills and attributes can be identified, the next question is whether they can be effectively taught thereby reducing noncompliance or permitting residents to carry larger patient loads.Hospitalists may be well suited to teach “efficiency” to housestaff. Hospitalists, both adult and pediatric, have reduced length of stay/costs between 10 – 15%. Hospitalist-educators are a growing group in academic centers who could focus on how best to teach efficiency to housestaff .
Our study has certain limitations. The current study relied on duty hour self-reporting, in a single-center and a single intern class. Duty hour reporting for the interns occurred at the end of each rotation creating the possibility of recall bias. Residents may underreport noncompliance . However, our sensitivity analysis, which attempted to mitigate the potential for this bias, demonstrated similar results. The sample size limited the number of potential contributing factors that could be evaluated. For example, we did not include attending characteristics in our model. The sample size also limited the power to detect individual intern factors associated with noncompliance.